Cytokine 142 (2021) 155499
Available online 30 March 2021
1043-4666/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Host biomarkers for monitoring therapeutic response in extrapulmonary tuberculosis
Atiqa Ambreen
a,b, Aasia Khaliq
c, Syed Zeeshan Haider Naqvi
b, Amna Tahir
d, Manal Mustafa
e, Safee Ullah Chaudhary
d, Shaper Mirza
d, Tehmina Mustafa
f,g,*aDepartment of Microbiology, Gulab Devi Hospital, Lahore, Pakistan
bInstitute of Molecular Biology and Biotechnology (IMBB), The University of Lahore, Defence Road Campus, Lahore, Pakistan
cSchool of Biological Sciences, University of the Punjab, Lahore, Pakistan
dBiomedical Informatics Research Laboratory, Department of Biology, Syed Babar Ali School of Science and Engineering, Lahore University of Management Sciences, Lahore, Pakistan
eLeverify (Business Intelligence Officer at Leverify), Lahore, Pakistan
fCentre for International Health, Department of Global Public Health and Primary Care, University of Bergen, Bergen, Norway
gDepartment of Thoracic Medicine, Haukeland University Hospital, Bergen, Norway
A R T I C L E I N F O Keywords:
Extra-pulmonary tuberculosis Lymphadenitis
Pleuritis
Response to treatment Inflammatory biomarkers
A B S T R A C T
Purpose: The aim of this study was to explore the utility of inflammatory biomarkers in the peripheral blood to predict response to treatment in extrapulmonary tuberculosis (EPTB).
Methods: A Luminex xMAP-based multiplex immunoassay was used to measure 40 inflammatory biomarkers in un-stimulated plasma of 91 EPTB patients (48 lymphadenitis, and 43 pleuritis) before and at 2 and 6 months of treatment.
Results: Overall a significant change was observed in 28 inflammatory biomarkers with treatment in EPTB pa- tients. However, MIG/CXCL9, IP-10/CXCL10, and CCL23 decreased in all patients’ groups with successful treatment at both time points. At 2 months, 29/64 (45%) patients responded partially while 35/64 (55%) showed complete regress. Among good responders, a higher number of biomarkers (16/40) reduced significantly as compared to partial responders (1/40). Almost half (14/29) of partial responders required longer treatment than 6 months to achieve satisfactory response. The levels of MIG, IP-10, MIF, CCL22 and CCL23 reduced significantly among 80, 74, 60, 71, 51% good responders, as compared to 52, 52, 52, 59, 52% partial responders, respectively. A biosignature, defined by a significant decrease in any one of these five biomarkers, corresponded with satisfactory response to treatment in 97% patients at 2 month and 99% patients at 6 months of treatment.
Conclusion: Change in inflammatory biomarkers correlates with treatment success. A five biomarker biosignature (MIG, IP-10, MIF, CCL22 and CCL23) could be used as an indicator of treatment success.
1. Introduction
Tuberculosis (TB) continues to be a major cause of morbidity and mortality in low and middle-income countries [1]. The global burden of the disease is estimated to be around 10 million as per World Health Organization and the prevalence of extrapulmonary TB (EPTB) is up to 24% of all notified TB cases [2]. The diagnosis of EPTB is challenging due to high variability in clinical presentation and difficulty in obtaining
a representative sample from the disease site for microbiological confirmation. Moreover, due to the paucibacillary nature of the disease, sensitivity of the routine microbiological tests is low [3,4]. This often leads physicians to diagnose EPTB on clinical basis followed by administration of anti-TB treatment without bacteriological confirma- tion [5]. Monitoring response early during treatment is, therefore, critical to reduce overtreatment, development of drug resistance, morbidity, and mortality. For patients that have bacteriologically
* Corresponding author at: Centre for International Health, Department of Global Public Health and Primary Care, University of Bergen, P.O. Box 7804, N-5020 Bergen, Norway.
E-mail addresses: atiqaambren@gmail.com (A. Ambreen), aasia.khaliq.pu@gmail.com (A. Khaliq), zeeshan.haider@imbb.uol.edu.pk (S.Z.H. Naqvi), AmnaTahir80@gmail.com (A. Tahir), manalmustafazia@gmail.com (M. Mustafa), safeeullah@lums.edu.pk (S.U. Chaudhary), shaper.mirza@lums.edu.pk (S. Mirza), tehmina.mustafa@uib.no (T. Mustafa).
Contents lists available at ScienceDirect
Cytokine
journal homepage: www.elsevier.com/locate/cytokine
https://doi.org/10.1016/j.cyto.2021.155499 Received 3 February 2021; Accepted 8 March 2021
confirmed EPTB, monitoring response during treatment is equally important as the treatment involves a prolonged administration of multiple antimicrobials and the decision on duration of treatment often depends on the patients’ response to prevent relapse. In case of smear- positive pulmonary TB, smear conversion is an important criterion for assessing response to treatment [6]. However, in case of EPTB, it is difficult to obtain repeated samples from the disease site during treat- ment, and response to treatment often relies on clinical criteria [7].
Currently, there is a lack of reliable objective criteria that can be used in the routine clinical practice for monitoring response to treatment in EPTB [7,8]. Some studies have explored the change in the levels of different immune biomarkers for this purpose [9,10]. However, most of the studies have small sample sizes [11,12], or use stimulation of im- mune cells [13–15]. Measurement of biomarkers in un-stimulated pa- tient’s plasma provides a direct means to observe the changes in biomarkers in response to treatment [16]. The aim of this study was to investigate change in the levels of inflammatory biomarkers in the un- stimulated plasma of EPTB patients during the treatment and their utility to accurately predict the response to treatment.
2. Material and methods
The study was conducted at Gulab Devi Hospital, a private not-for- profit tertiary care hospital located in Lahore, Pakistan, and provides specialized TB care. Presumptive and diagnosed TB patients are referred from various districts for consultation and/or treatment at Gulab Devi Hospital Lahore. Patients of all ages with presumptive EPTB attending outpatient clinics were enrolled from April 2016 to August 2017. All patients received standard anti-TB treatment and followed up till the satisfactory response to treatment. Blood samples (5 ml) were collected at before initiation of anti-TB treatment (baseline) and at 2 and 6 months after treatment, centrifuged for 10 min at 1000g, plasma collected and frozen at − 20 ◦C for a few months and then shifted to − 80 ◦C until use.
2.1. Laboratory methods
For patients with enlarged lymph nodes, an excision biopsy was performed, and the sample was sent for histopathology and microbio- logical examination. For patients with pleural effusions, aspirated fluid was sent for cytology and microbiological workup. The specimens were processed for smear examination, Xpert MTB/RIF assay (Xpert), and culture [17]. Auramine O-stained smears were examined using a light- emitting diode fluorescence microscope [18]. Xpert was performed ac- cording to manufacturer’s protocols [19]. Two slopes of Lowenstein- Jensen medium and one Mycobacteria Growth Indicator Tube (MGITTM 960TM; Becton Dickinson, Sparks, MD, USA) were inoculated for culture [17].
2.2. Inflammatory biomarkers detection through multiplex microbead immunoassay
Biorad 40 plex Bio-PlexProTMHuman Chemokine Panel (Table 1), was used on Luminex® xMAP™ to detect cytokines/chemokines from the plasma (Referred to as inflammatory biomarkers in the text). Frozen plasma samples were thawed, mixed by vortexing, and centrifuged for 10 min at 10,000g to remove particulates before the assays were per- formed. Plasma samples were analyzed in duplicates in the first exper- iment and singlets in the succeeding experiments as inter-assay variability was in the acceptable range. Blanks and standards were run in duplicates in all experiments. Assays were performed as per manu- facturer’s instructions (BioRad, Hercules, CA). Briefly, after pre-wetting the plates, 50 μl of 1x beads were added to wells, plates were washed twice and 50 ul of standards, controls, and samples were added to the respective wells. After one hour’s incubation on a shaker at room tem- perature, plates were washed 3 times and 25 ul of detection antibodies were added to each well. After an incubation of 30 min at room tem- perature and washing thrice, 50 ul of streptavidin-E was added to each well. Plates were incubated for another 10 min on the shaker at room temperature and after three steps of washings, re-suspended with 125 ul of assay buffer. Plates were read with a Luminex instrument (Luminex 200, Austin Luminex, USA). Data was analyzed using MILLIPLEX Ana- lyst 5.1 software (Merck Millipore Darmstadt, Germany), as per the manufacturer’s instructions.
2.3. Case definition
Using a combination of clinical, radiological, and laboratory find- ings, cases were defined as confirmed and probable EPTB cases. A confirmed case was defined based on the bacteriological confirmation either on culture or Xpert. A probable TB pleuritis case was defined if the symptoms and findings were consistent with TB pleuritis (lymphocytosis and fluid protein level more than 3 g/dl or plasma adenosine deaminase levels more than 16 U/L or concomitant pulmonary TB suggested by positive acid-fast bacilli smear and/or chest radiograph) and with good response to anti-TB treatment at 2–3 months and/or end of treatment. A probable TB lymphadenitis case was defined if the symptoms, clinical findings, and histopathology were consistent with TB lymphadenitis and with good response to anti-TB treatment at 2–3 months and /or end of the treatment.
2.4. Response to treatment
The response to treatment was considered as good if two of these three criteria were fulfilled, 1) regression of symptoms, 2) regression of local signs of disease; regression of lymph nodes among lymphadenitis cases and regression of pleural effusion assessed by ultrasound among the pleuritis cases, 3) weight gain.
2.5. Statistical analysis
International Business Machine (IBM) – Statistical Package for Social Sciences (SPSS) version 23 and R studio were used for data analysis.
Data were evaluated for normality using Shapiro-Wilk and Kolmogorov- Smirnov tests. For normally distributed data, paired t-test was used, otherwise non-parametric Wilcoxon signed rank test was employed for analysis of data. Chi-square test was done for categorical data. A p-value
< 0.05 was considered statistically significant. Furthermore, Linear Discriminant Analysis was used to classify cytokines at baseline and at two points after treatment. Ratio (or pirate) plots with 95% highest density interval were used to visualize the ratio of change in the biomarker levels for individual patients.
Different biosignatures were synthesized by making a combination of the inflammatory biomarkers which showed statistically significant change in the median levels at two timepoints during treatment as Table 1
Cytokine and chemokine panel used on plasma samples of the TB pleuritis and lymphadenitis patients.
Pro-inflammatory
cytokines Interferon-gamma (IFN-γ), Tumor necrosis factor Alpha (TNF-α), IL-1β, IL-6, IL-8, IL-16, MIF
Anti-inflammatory
cytokines IL-4, IL-10
Chemokines CCL 6Ckine/CCL21, CTACK/CCL27, Eotaxin/CCL11, Eotaxin-2/CCL24, Eotaxin-3/CCL26, 309/CCL1, MCP-1/CCL2, MCP-2/CCL8, MCP-3/CCL7, MCP- 4/CCL13, MDC/CCL22, TECK/CCL25, TARC/
CCL1, MIP-3β/CCL19, MIP-3α /CCL20, MPIF-1/
CCL23, MIP-1δ/CCL15
CXCL BCA-1/CXCL13, ENA-78/CXCL5, GCP-2/CXCL6, Gro-α/CXCL1, Gro-β/CXCL2, IL-8/CXCL8, IP-10/
CXCL10, I-TAC/CXCL11, MIG/CXCL9, SDF- 1α+β/CXCL12, SCYB16/CXCL16 CX3CL Fractalkine/CX3CL1
Growth factors Granulocyte-macrophage colony-stimulating factor (GM- CSF), IL-2
compared to the baseline, and each biomarker showed ≥20% change from the baseline in an individual patient. A software library, Python And Data Analysis (Pandas), was used for basic data manipulation and all the possible combinations of members of the given biomarkers set were computed. The biomarker combinations covering the greatest number of patients and the least number of biomarkers were selected.
3. Results
3.1. Patients characteristics
Fig. 1 shows the patients included in the study. A total of 671 pre- sumptive EPTB cases were investigated during the study period. Among them, 364 were registered at Gulab Devi Hospital. A total of 91 patients were included in the study. The demographic and clinical characteristics of study participants are shown in Table 2. There were 48 TB lymph- adenitis and 43 TB pleuritis cases. Using a composite reference standard, 51 patients were classified as confirmed, and 40 as probable EPTB cases.
The median age was 20 years for lymphadenitis and 25 years for pleuritis patients. There was a preponderance of females (75%) among lymphadenitis patients while the majority (70%) of patients presenting with pleuritis were males. None of these patients was positive for human immunodeficiency virus. One lymphadenitis and 3 pleuritis patients had a history of having diabetes. All other patients had random blood sugar levels less than 200 mg/dl. At the second month of treatment 20/48 (42%) lymphadenitis and 32/43 (74%) pleuritis patients showed satis- factory response to treatment with improvement of clinical and sub- jective criteria. Rest of the patients showed clinical improvement but not complete resolution of signs and symptoms. At 6 months of treatment, 44/48 lymphadenitis and 41/43 pleuritis patients turned up for follow- up. Treatment was extended for 20 patients (15 lymphadenitis and 5 pleuritis), 16 of these 20 patients showed partial response at 2 months of treatment. Clinical improvement was recorded in all patients at the end of the treatment.
3.2. Inflammatory biomarker profile change with treatment in TB lymphadenitis
Fig. 2 shows the 16 biomarkers that changed significantly with treatment as compared to the baseline. After 2 months of treatment, there was a significant decrease in the plasma levels of MIG (p =.007), IP-10 (p =.025), CXCL2 (p =.042), CCL8 (p =.048), CCL22 (p =.003), and CCL23 (p =.009), and an increase in the plasma levels of TNF-α (p =
EPTB cases registered for
treatment = 364
TB lymphadenitis = 48
Bacteriology confirmed cases = 38 Probable EPTB cases = 10
TB pleuritis = 43 Included in the study = 94
Not included in the study = 270 (Sample either not collected at 0 M or if present at 0 M missing at both 2 and 6 M)
Bacteriology confirmed cases = 13 Probable EPTB cases = 30
Patients with sample at 0 M + 2 M + 6 M =30 0 M + 2 M = 8 0 M + 6 M = 10
Patients with sample at 0 M + 2 M + 6 M = 18 0 M + 2 M = 8 0 M + 6 M = 17 Non-Tuberculous Mycobacteria
(Mycobacterium Fortuitum) = 3 Multiplex analysis performed = 91 Presumptive EPTB cases investigated
(April 2016- August 2017) = 671
Fig. 1. Flow chart showing patients included in the study and the number of plasma samples obtained at different time points during treatment. Abbreviations, TB:
tuberculosis, EPTB: extrapulmonary tuberculosis, M: Month of the treatment.
Table 2
Demographic and clinical characteristics of extrapulmonary tuberculosis patients.
Patient Characteristics TB lymphadenitis N
=48 TB pleuritis N
=43 Age in years, median, (range) 20 (11–72) 25 (15–70) Sex, n (%)
Male 12 (25) 30 (70)
Female 36 (75) 13 (30)
HIV status, n (%)
Positive 0 (0) 0 (0)
Negative 48 (100) 43 (100)
History of Diabetes, n/N (%)
Yes 1/45 (2) 3/37 (8)
No 44/45 (98) 34/37 (92)
NA 3/48 (6) 6/43 (14)
Patient Categorization, n/N (%)
Confirmed TB 38/48 (79) 13/43 (30)
Culture positive 11/38 (29) 10/13 (77)
Xpert Positive 3/38 (8) 1/13 (8)
Both positive 24/38 (63) 2/13 (15)
Probable TB 10/48 (21) 30/43 (70)
Clinical response at 2 M of treatment, n/N (%)
Responders
Partial responders 20/48 (42)
28/48 (58) 32/43 (74) 11/43 (26) Clinical response at 6 M/end* of
treatment, n/N (%)
Responders 44/44 (100) 41/41 (100)
N: Total number, n: number, %: percentage, NA : no information available TB:
tuberculosis, M: month.
*Treatment was extended for 15 lymphadenitis and 5 pleuritis patients beyond 6 months.
Fig. 2. Box plots showing plasma levels of inflammatory biomarkers in lymphadenitis patients at baseline, 2, and 6 months of treatment. Biomarkers that changed significantly with treatment are shown. The Wilcoxon signed rank test was used to compare biomarkers expression at different time points. A p-value <0.05 was considered significant. Boxes represent the median and interquartile range, and the whisker represents minimum/maximum values. Outliers are shown by a broken axis. n =number of patients.
Fig. 3. Linear discriminant analysis (LDA) at baseline and at 2 & 6 months of treatment. a: LDA for tuberculous lymphadenitis patients (n =30). LDA gave significant classification (Wilk Lambda =0.016) with a membership of 100% at 6 months. b: LDA for tuberculous pleuritis patients (n =18). Although statistical significance was not obtained (Wilk Lambda =0.119, with a membership of 62.5% at 6 months) due to small sample size, there is a visible trend of separate clustering at 0, 2, and 6 months.
.021), CCL2 (p =.004), CCL3 (p =.005), CCL13 (p =.045) and CCL26 (p
=.008). After 6 months of treatment, levels of MIG (p <.001), IP-10 (p
=.024), MIF (p =.044), IL-1β (p =.004), CXCL11 (p =.004), and CCL23 (p = .003) decreased significantly, while levels of CCL2 (p = .013), CCL13 (p =.002), CCL26 (p =.028), CXCL1 (p =.007), and CXCL12 (p
=.022) increased significantly as compared to the baseline.
Linear discriminant analysis (Fig. 3a) gave a significant classification of patients with respect to their above-mentioned inflammatory bio- markers’ levels at baseline, 2, and 6 months, further depicting signifi- cant change in the levels of inflammatory biomarkers after treatment at both time points, with a membership of 100% at 6 months.
Fig. 4 shows the ratio of change in the levels of above-mentioned biomarkers in individual patients at both time points showing wide
variation between individual patients. All the above-mentioned bio- markers did not change in all patients. Table 3 shows the proportions of patients in which the individual biomarkers showed ≥20% change from the baseline in an individual patient. The magnitude of change (fold change) also varied between biomarkers ranging from 2- to12-fold change (Table 3).
3.3. Inflammatory biomarker profile change with treatment in TB pleuritis Fig. 5 shows the 24 inflammatory biomarkers that changed signifi- cantly with treatment as compared to the baseline. After two months of treatment, a significant decrease in plasma levels of MIG (p =.002), IP- 10 (p <.001), IFN-γ (p =.004), IL-4 (p =.022), CCL1 (p <.001), CCL8 (p Fig. 4.Ratio/pirate plots to visualize the decrease or increase in plasma levels of in- flammatory biomarkers in the individual tuberculous (TB) lymphadenitis patients.
Only those inflammatory biomarkers that changed significantly with treatment are shown. a: Ratio of plasma level of inflam- matory biomarkers at 2 month of treatment to their levels at baseline (n =38) *Outliers:
CCL13: 7 & 159, CCL2: 6; 14; 16 & 27, CCL22: 20, CCL23: 14, CCL3: 6; 8; 14; 22; 22;
26; 27; 76 & 209, CCL8: 1300, CXCL2: 8; 10
& 139, CCL26: 7; 8; 8; 10; 12; 22; 25; 164;
241 & 351, IP10: 6 & 8, TNF-α: 6; 9; 9 & 40.
b: Ratio of plasma levels of inflammatory biomarkers at 6 months of treatment to their levels at the start of treatment (n = 40)
*Outliers: CCL13: 6; 7; 7; 9 & 9, CCL2: 8 &
166, CCL23: 7 &15, CCL26: 6, 8, 9, 9, 10, 14, 42, 76, 104, 164 & 176, CXCL1: 6, 6, 8, 9, 12, 15, 17 & 551, CXCL12: 6, 6, 7, 8, 8, 26 &
251, IP10: 6, MIF: 10, 18, 31, 80 & 116.
<.001), CCL20 (p =.005), CCL22 (p =.002), CCL23 (p =.001), CCL24 (p =.012), CXCL2 (p <.001), CXCL11 (p =.001), and CX3CL1 (p = .016), and an increase in levels of IL-8 (p =.012), CCL26 (p =.028), and CCL13 (p =.034) was observed. After 6 months of treatment, a signif- icant decrease was seen in the levels of 18 inflammatory biomarkers as compared to the baseline. These included MIG (p <.001), IP-10 (p <
.001), IFN-γ (p =.002), MIF (p =.013), IL-4 (p =.035), IL-6 (p <.001), CCL1 (p <.001), CCL3 (p =.038), CCL8 (p =.003), CCL15 (p =.041), CCL17 (p =.013), CCL19 (p =.036), CCL20 (p =.011), CCL23 (p <
.001), CXCL2 (p =.027), CXCL6 (p =.021), CXCL11 (p <.001), and CX3CL1 (p =.023). However, a significant increase was seen for only CCL2 (p =.012).
Linear discriminant analysis as shown in Fig. 3b gave a significant classification of patients with respect to their above-mentioned inflam- matory biomarkers’ levels at 2 months but not at 6 months. However, at 6 months, there was a visible trend depicted by clustering when compared with the baseline.
Fig. 6 shows the ratio of change in the levels of above-mentioned biomarkers in individual patients at both time points. As in lymphade- nitis, a wide variation was seen between individual patients. All the above-mentioned biomarkers did not change in all patients. Table 3 shows the proportions of patients in which the individual biomarkers showed ≥20% change from the baseline in an individual patient. The magnitude of change (fold change) also varied between biomarkers ranging from 2- to12-fold change (Table 3). A biomarker with higher fold change implies its robustness as an indicator of treatment response.
3.4. Common inflammatory biomarkers changing in both TB pleuritis and lymphadenitis
Fig. 7 shows the common biomarkers that changed significantly in both patient groups at both time points. After two months of treatment, MIG, IP-10, CXCL2, CCL8, CCL22, and CCL23 decreased, while CCL13, and CCL26 levels increased in both groups. At 6 months, levels of MIG, MIF, IP-10, CCL23, and CXCL11 decreased and the level of one biomarker (CCL2) increased in both groups of patients.
Table 3
Proportion of tuberculous lymphadenitis and pleuritis patients showing ≥20%
change in the levels of biomarkers in response to treatment and the magnitude of change (fold change) in median plasma levels.
Biomarkers TB lymphadenitis TB pleuritis
0-2ф 0-6¥ 0–2 ф 0-6¥
n/N (%) FC
↓/↑ n/N
(%) FC
↓/↑ n/N
(%) FC
↓/↑ n/N
(%) FC
↓/↑
MIG 22/
38 (58)
3↓ 31/
40 (76)
6↓ 21/
26 (81)
2↓ 29/
35 (83) 7↓
IP-10 21/
38 (55)
3↓ 22/
40 (55)
3↓ 20/
26 (77)
3↓ 27/
35 (77) 3↓
CCL23 19/
38 (50)
2↓ 21/
39* (54)
3↓ 14/
26 (54)
3↓ 26/
35 (74) 2↓
CXCL11 28/
40 (70)
4↓ 19/
26 (73)
2↓ 29/
35 (83) 6↓
CCL22 24/
38 (63)
2↓ 18/
26 (69) 2↓
CCL2 23/
38 (61)
3↑ 25/
40 (63)
3↑ 19/
35 (54) 2↑
CCL26 22/
38 (58)
5↑ 24/
40 (60)
5↑ 18/
26 (69) 4↑
CCL13 21/
38 (55)
3↑ 30/
40 (75)
2↑ 13/
26 (50) 3↑
CXCL2 20/
38 (53)
5↓ 22/
26 (85)
7↓ 21/
35 (60) 5↓
CCL8 20/
38 (53)
3↓ 22/
26 (85)
3↓ 21/
35 (60) 2↓
CCL3 23/
38 (61)
7↑ 23/
35 (66) 2↓
TNF-α 24/
38 (63) 3↑
IFN-γ 19/
26 (73)
2↓ 19/
35 (54) 3↓
CCL1 21/
26 (81)
2↓ 23/
35 (66) 2↓
IL-4 17/
24* (71)
3↓ 19/
32* (54) 8↓
CXCL12 26/
40 (65) 3↑
CXCL1 25/
39* (64) 4↑
MIF 27/
40 (68)
12↓ 25/
35 (71) 5↓
IL-1β 27/
40 (68) 3↓
CCL20 18/
26 (69)
3↓ 24/
35 (69) 4↓
CX3CL1 18/
26 (69)
3↓ 19/
35 (54) 2↓
CCL24 19/
26 (73) 12↓
IL-8 12↑
Table 3 (continued)
Biomarkers TB lymphadenitis TB pleuritis
0-2ф 0-6¥ 0–2 ф 0-6¥
n/N (%) FC
↓/↑ n/N
(%) FC
↓/↑ n/N
(%) FC
↓/↑ n/N
(%) FC
↓/↑
16/ 26 (62)
CXCL6 19/
35 (54) 5↓
CCL15 15/
35 (43) 2↓
CCL17 25/
35 (71) 3↓
CCL19 21/
35 (60) 3↓
IL-6 29/
35 (83) 3↓
FC: fold change, n: number of patients showing significant change, N: total number of patients, ↓: decrease, ↑: increase.
ф From baseline to 2 months after treatment.
¥ From baseline to 6 months after treatment.
* Valid available values.
3.5. Biosignature predicting response to treatment in all patients
Biosignatures were synthesized by different possible combinations of biomarkers on the condition that change in any one of the biomarkers in the biosignature would predict response to treatment. The goal was to have a biosignature with minimum number of biomarkers changing in maximum number of patients and in both forms of TB. When all significantly changed biomarkers were used, the number of possible combinations became too high. We, therefore, selected five biomarkers, MIG, IP-10, MIF, CCL22, CCL23, based on their high baseline levels, and their change in higher proportion of patients. All possible combinations with these five biomarkers are shown in (Online Resource 1) Supple- mentary Tables 1, 2, & 3. Table 4 shows the selected biosignatures and their sensitivity to predict response to treatment at 2 and 6 months of the treatment. A biosignature (MIG + IP-10 +MIF +CCL22 +CCL23), could predict response to treatment in 97% patients at 2 months and 99% patients at 6 months of treatment.
3.6. Change in plasma inflammatory biomarkers correlates with early clinical response during treatment
After 2 months of treatment, 35/64 (55%) patients (16 lymphade- nitis and 19 pleuritis) showed good clinical response (responders), while 29/64 (45%) patients (22 lymphadenitis and 7 pleuritis) improved clinically but clinical signs did not settle completely (partial re- sponders). Fig. 8 shows the levels of biomarkers among responders and partial responders. A total of 16 inflammatory biomarkers decreased significantly in responders while only one decreased in partial re- sponders, indicating that a decrease in plasma levels of these biomarkers at 2 months correlates with good clinical response. Furthermore, re- sponders showed significant decline in four of the biomarkers (MIG, IP- 10, CCL22, CCL23) included in the biosignature as compared to the partial responders who showed significant change in only one (CCL23).
Extension of treatment was required for 20 patients and 16/20 were
partial responders. Plasma samples were available for 14/16 patients at 2 months. This implies that these biomarkers can also be used to predict which patients would require prolonged treatment beyond 6 months.
Although decrease in MIF levels did not reach statistical significance at 2 months, ≥20% decrease was seen in 60% responders and 52% partial responders (Fig. 8).
4. Discussion
In this prospective cohort study, we have shown that plasma levels of several inflammatory biomarkers change with treatment. The patients’ response to infection and treatment would depend on a variety of host and bacteriological factors and is expected to vary among individuals and between different disease sites. A single inflammatory biomarker is, therefore, not expected to give a satisfactory response in all patient categories, while a combination of biomarkers can predict response to treatment with reasonable certainty in many patients as shown in our study. Individually, the levels of IP-10, MIG, and CCL23 changed significantly in majority of TB lymphadenitis and TB pleuritis patients at both 2 and 6 months after treatment. A combination of five inflamma- tory biomarkers (MIG, IP-10, MIF, CCL22 and CCL23) could predict response to treatment in 97% of our study patients at 2 and 99% at 6 months after treatment.
Several studies have shown that MIG, MIF, IP-10 and CCL22 plasma levels increase in active TB and decline with successful treatment and have been proposed as surrogate markers for the evaluation of treatment response in pulmonary and EPTB. However, the main focus of these studies has been pulmonary TB with relatively few EPTB cases [20–24].
Few other studies have shown various combinations of biomarkers in plasma [9] or saliva [25] to assess response to treatment in pulmonary TB. To our knowledge this is the first study to evaluate the role of 40 inflammatory biomarkers in >90 EPTB patients at two time points after treatment.
Although several biomarkers changed with treatment, and many Fig. 5. Box plots showing changes in plasma levels of inflammatory biomarkers in tuberculous pleuritis patients at baseline and 2, and 6 months of treatment.
Biomarkers that changed significantly with treatment are shown. The Wilcoxon signed rank test was used to compare biomarkers expression at different time points.
A p-value <0.05 was considered significant. Boxes represent the median and interquartile range, and the whisker shows minimum/maximum values. Outliers are shown by a broken axis. n =number of patients.
combinations gave good sensitivity in assessing response to treatment, we selected a biosignature comprising of the biomarkers with, i) a persistent trend (downward or upward) throughout the treatment, ii) high plasma levels, iii) high fold change and iv) giving coverage to maximum number of patients. The relatively high plasma levels make them potential candidates for detection by less sensitive techniques than multiplex assay, and development into tests for routine clinical use.
Furthermore, their higher levels in plasma imply that these biomarkers would be detectable in the dried blood spots as shown by previous studies on IP-10 on dry blood spots [26,27]. This opens the possibility of developing a point-of-care test based on these biomarkers. Dried blood spots are easy to make and can be transported at ambient temperature to a reference laboratory.
After 2 months of treatment, patients that responded well to treat- ment showed marked change in biomarkers levels as compared to the partial responders, indicating that significant change in plasma levels of inflammatory biomarkers correlates with good clinical response and can be used to predict response to treatment during initial months of the therapy. Since majority of the partial responders at 2 months required prolongation of treatment, these biomarkers can also be used to predict the need for prolonged duration of treatment to achieve satisfactory response. Previous studies have shown that inflammatory markers can
be used for the evaluation of the early treatment response [20,28].
All patients in our cohort showed good response to treatment at the end of treatment. This is not in agreement with our clinical experience and experience from other cohorts [29]. Usually, some of the patients started on anti-TB treatment on clinical suspicion do not show satis- factory response to treatment. This could be due to the bias introduced by the study design, as the attending physician knew about the ongoing study and was extra careful in the selection of patients before the start of anti-TB treatment. Whereas in routine practice more liberal prescription of anti-TB treatment is done in presumptive EPTB cases due to lack of a reliable diagnostic test, leading to overdiagnosis and overtreatment.
Diabetes mellitus has been documented as a risk factor for TB [30,31], and the prevalence of diabetes in Pakistan is reported to be around 26% [32]. However, only 4 patients (5%) in our cohort were diabetics. This is in agreement with our previous studies on EPTB where the prevalence of diabetes has been shown to be low among EPTB pa- tients in Zanzibar (2%) [33], and India (2%) [34]. We also reported low prevalence of diabetes among pulmonary TB patients from Pakistan (5%) [35]. In agreement, another study from India has also reported low prevalence (5.4%) of diabetes among 37 EPTB patients [36], implying that diabetes might not be a risk factor for EPTB. However, this study was not designed to study the correlation between diabetes and EPTB, Fig. 6. Ratio/pirate plots to visualize the decrease or increase in plasma levels of in- flammatory biomarkers in the individual tuberculous pleuritis patients. Only those inflammatory biomarkers that changed significantly with treatment are shown. a:
Ratio of plasma level of inflammatory bio- markers at 2 months of treatment to their levels at the start of treatment (n = 26).
*Outliers: CCL13: 11 & 13, CCL26: 6; 9; 9;
16; 21; 28; 29& 37, CXCL2: 11, IL8: 7; 8; 11;
12; 25; 30; 43; 44 & 77. IL4: 130. b: Ratio of plasma levels of inflammatory biomarkers at 6 months of treatment to their levels at the start of treatment (n =35). *Outliers: CCL19:
14, CCL2: 6, CCL20: 12 & 19, CCL23: 7, CCL3: 10 & 36, IL4: 36, IL6: 47, MIF: 15; 16;
19 & 21.
and further studies are needed to address this.
Our study has few weaknesses, i) all patients included in the study did not have bacteriologically confirmed TB. However, this reflects the situation in routine clinical practice due to the paucibacillary nature of the EPTB, making our results more generalized for a heterogeneous group of patients. ii) Relapse rate was not studied as patients were not followed-up after completion of treatment. iii) Our sample size was not
as large as we had anticipated, as many EPTB patients registered for anti- TB treatment refused to give blood samples for research purposes or were lost to follow-up. The results from this study need to be validated in larger patient populations as well as in different epidemiological settings.
Fig. 7. Venn diagram showing common inflam- matory biomarkers that changed significantly at different time points in tuberculous (TB) lymph- adenitis and pleuritis patients, n: number of pa- tients, 0 M–2 M: Significant change at 2 months of the treatment as compared to baseline, 0 M–6 M: Significant change at 6 months of the treat- ment as compared to baseline, ↑: significant in- crease in plasma levels with treatment, ↓: significant decrease in plasma levels with treat- ment, *CCL3 levels increased significantly at 2 M in lymphadenitis patients, whereas it decreased significantly in pleuritis patients.
Table 4
The proportion of EPTB patients showing significant change in the levels of five biomarkers constituting the biosignatures at 2 and 6 months of treatment, and the sensitivity of the biosignatures to predict response to treatment.
Immune Biomarkers
(Median at 0–2-6 months pg/ml) All samples (0 M–2 M) N =64 n (%)
All samples (0 M−6 M) N =75 n (%)
Responders (0 M−2 M) N =35 n (%)
Partial responders (0 M−2 M) N =29 n (%)
P-value**
MIG (2276 – 1418 – 568) 43 (67) 60 (80) 28(80) 15 (52) .016
IP-10
(481 – 269 – 233) 41 (64) 49 (65) 26 (74) 15 (52) .061
MIF (13990 – 9407 – 5773) 36 (56) 52 (69) 21 (60) 15 (52) .506
CCL22 (884 – 626 – 670)
42 (66) 32 (43) 25 (71) 17 (59) .283
CCL23
(468 – 327 – 301) 33 (52) 47/74* (64) 18 (51) 15 (52) .910
MIG +IP-10 51 (80) 66 (88) 32 (91) 19 (66) .010
MIG +CCL23 51 (80) 70 (93) 31 (89) 20 (69) .052
MIG +MIF 54 (84) 68 (91) 31 (89) 23 (79) .310
MIG +CCL22 54 (84) 64 (85) 33 (94) 21 (72) .016
MIG +IP-10 +CCL22 57 (89) 67 (89) 34 (97) 23 (79) .023
MIG +MIF +CCL23 59 (92) 74 (99) 33 (94) 26 (90) .492
MIG +MIF +IP-10 58 (91) 72 (96) 34 (97) 24 (83) .049
MIG +IP-10 +CCL23 54 (84) 70 (93) 32 (91) 22 (76) .088
IP-10 +MIF +CCL23 59 (92) 70 (93) 33 (94) 26 (90) .492
MIG +IP-10 +MIF +CCL22 61 (95) 72 (96) 35 (100) 26 (90) .051
MIG +MIF +IP-10 +CCL23 60 (94) 74 (99) 34 (97) 26 (90) .218
MIG +MIF +IP-10 +CCL22 +CCL23 62 (97) 74 (99) 35 (100) 27 (93) .114
n: number of patients showing significant change, N: total number of patients, M: month of treatment.
*Valid available values.
**Chi-square test was done to see difference of biomarkers coverage among responders and partial responders at 2 months of treatment. A p-value <0.05 was considered statistically significant
5. Conclusion
A biosignature including MIG, IP-10, MIF, CCL22, and CCL23 was reliable in predicting response to treatment in EPTB at 2 and 6 months after standard anti-TB treatment in our study cohort. Relatively high plasma levels of biomarkers included in the biosignature imply the possibility of developing it further into a test for routine use by using less sensitive ELISA method and using less invasive sampling as dry blood spots.
Funding
This work was partly supported by the Research Council of Norway through the Global Health and Vaccination Program (project number 234457). This project is part of the EDCTP2 programme supported by the European Union.
7. Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
8. Author’s contributions
Concept and study design: TM, AA. Acquisition of funds for the study:
TM. Performing the experiments: AA, AK, SM. Analysis and interpreta- tion of data: TM, AA, AT, SUC, SM, MM. Drafting and revising the
manuscript: AA, TM, SM, SUC, AK, AT, MM, SZHN. All authors read and approved the final manuscript.
Ethics declarations
The study was approved by the Institutional Review Board, Al. Aleem Medical College & Gulab Devi Educational Complex Lahore (GDEC/18- 322) and Regional Committee for Medical and Health Research Ethics, Western-Norway (2018/2392/REK vest). All study participants pro- vided informed consent.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
Authors would like to thank Dr Nauman Safdar from the Social and Health Inequalities Network (SHINe), and Dr Muhammad Jamil from chest medicine department at Gulab Devi Hospital for helping in recruiting and examination of patients. We are thankful to Mr Hafeez Ullah and Miss Nadia Ameer Hamza for their contribution in sample handling in the laboratory.
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.cyto.2021.155499.
Fig. 8. a: Venn diagram showing inflammatory biomarkers that changed significantly in lymphadenitis and pleuritis patients showing good clinical response (re- sponders), and partial clinical response (partial responders). A paired t-test (p <0.05) revealed that a total of 16 inflammatory biomarkers decreased significantly in responders while only one decreased in partial responders. *Inflammatory biomarkers included in the biosignature ↑: significant increase, ↓: significant decrease, M:
month of treatment. b-r: Mean concentrations of inflammatory biomarkers that changed significantly with treatment in responders n =23 (continuous line) and partial responders n =25 (dotted line). The vertical bars show standard error of the mean.
References
[1] N.R. Gandhi, P. Nunn, K. Dheda, H.S. Schaaf, M. Zignol, D. Van Soolingen, et al., Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control of tuberculosis, Lancet 375 (9728) (2010) 1830–1843.
[2] Y.K. Yong, H.Y. Tan, A. Saeidi, W.F. Wong, R. Vignesh, V. Velu, et al., Immune biomarkers for diagnosis and treatment monitoring of tuberculosis: current developments and future prospects, Front. Microbiol. 10 (2019).
[3] D. Hillemann, S. Rüsch-Gerdes, C. Boehme, E. Richter, Rapid molecular detection of extrapulmonary tuberculosis by the automated GeneXpert MTB/RIF system, J. Clin. Microbiol. 49 (4) (2011) 1202–1205.
[4] B. Malbruny, G. Le Marrec, K. Courageux, R. Leclercq, V. Cattoir, Rapid and efficient detection of Mycobacterium tuberculosis in respiratory and non- respiratory samples, Int. J. Tuberc. Lung Dis. 15 (4) (2011) 553–555.
[5] S. Tahseen, F.M. Khanzada, A.Q. Baloch, Q. Abbas, M.M. Bhutto, A.W. Alizai, et al., Extrapulmonary tuberculosis in Pakistan-A nation-wide multicenter retrospective study, PLoS ONE 15 (4) (2020), e0232134.
[6] World Health Organization Stop TB Initiative, Treatment of tuberculosis:
guidelines fourth edition. World Health Organization, 2010 [cited 2020 15 June 2020]. Available from: <https://apps.who.int/iris/bitstream/handle/10665/
44165/9789241547833_eng.pdf>.
[7] J.Y. Lee, Diagnosis and treatment of extrapulmonary tuberculosis, Tuberc. Respir.
Dis. 78 (2) (2015) 47–55.
[8] N. Rockwood, E. du Bruyn, T. Morris, R.J. Wilkinson, Assessment of treatment response in tuberculosis, Expert Rev. Respir. Med. 10 (6) (2016) 643–654.
[9] R. Jacobs, S. Malherbe, A.G. Loxton, K. Stanley, G. Van Der Spuy, G. Walzl, et al., Identification of novel host biomarkers in plasma as candidates for the immunodiagnosis of tuberculosis disease and monitoring of tuberculosis treatment response, Oncotarget 7 (36) (2016) 57581.
[10] V. Clifford, C. Zufferey, A. Street, J. Denholm, M. Tebruegge, N. Curtis, Cytokines for monitoring anti-tuberculous therapy: a systematic review, Tuberculosis 95 (3) (2015) 217–228.
[11] K. Tonby, M. Ruhwald, D. Kvale, A.M. Dyrhol-Riise, IP-10 measured by Dry Plasma Spots as biomarker for therapy responses in Mycobacterium Tuberculosis infection, Sci. Rep. 5 (2015) 9223.
[12] Y.-G. Hur, Y.A. Kang, S.-H. Jang, J.Y. Hong, A. Kim, S.A. Lee, et al., Adjunctive biomarkers for improving diagnosis of tuberculosis and monitoring therapeutic effects, J. Infect. 70 (4) (2015) 346–355.
[13] D. Kassa, W. de Jager, G. Gebremichael, Y. Alemayehu, L. Ran, J. Fransen, et al., The effect of HIV coinfection, HAART and TB treatment on cytokine/chemokine responses to Mycobacterium tuberculosis (Mtb) antigens in active TB patients and latently Mtb infected individuals, Tuberculosis 96 (2016) 131–140.
[14] A. Mihret, M. Abebe, Cytokines and chemokines as biomarkers of tuberculosis, J. Mycobac. Dis. 3 (2) (2013) 128.
[15] B.S.A. Kabeer, A. Raja, B. Raman, S. Thangaraj, M. Leportier, G. Ippolito, et al., IP- 10 response to RD1 antigens might be a useful biomarker for monitoring tuberculosis therapy, BMC Infect. Dis. 11 (1) (2011) 135.
[16] R. Ravindran, V.V. Krishnan, A. Khanum, P.A. Luciw, I.H. Khan, Exploratory study on plasma immunomodulator and antibody profiles in tuberculosis patients, Clin.
Vaccine Immunol. 20 (8) (2013) 1283–1290.
[17] K. Stinson, K. Eisenach, S. Kayes, M. Matsumoto, S. Siddiqi, S. Nakashima, et al., Mycobacteriology laboratory manual. Global laboratory initiative, a working group of the stop TB partnership, WHO, Geneva, 2014.
[18] R. Lumb, A. Van Deun, I. Bastian, M. Fitz-Gerald, The handbook. Global laboratory initiative. Laboratory diagnosis of tuberculosis by sputum microscopy, Switzerland: Geneva, 2013 [24 April 2020]; Global. Available from: <http://www.
stoptb.org/wg/gli/assets/documents/TB%20microscopy%20handbookfinal.pdf>.
[19] World Health Organization, Xpert MTB/RIF implementation manual, 2014.
Switzerland: Geneva, 2014 [24 April 2020]. Available from: <https://www.who.
int/tb/publications/xpert_implem_manual/en/>.
[20] W.Y. Chung, D. Yoon, K.S. Lee, Y.J. Jung, Y.S. Kim, S.S. Sheen, et al., The usefulness of serum CXCR3 ligands for evaluating the early treatment response in tuberculosis: a longitudinal cohort study, Medicine 95 (17) (2016).
[21] Q. Yang, Y. Cai, W. Zhao, F. Wu, M. Zhang, K. Luo, et al., IP-10 and MIG are compartmentalized at the site of disease during pleural and meningeal tuberculosis and are decreased after antituberculosis treatment, Clin. Vaccine Immunol. 21 (12) (2014) 1635–1644.
[22] Z-b Shang, J. Wang, S-g Kuai, Y-y Zhang, Q-f Ou, H. Pei, et al., Serum macrophage migration inhibitory factor as a biomarker of active pulmonary tuberculosis, Ann.
Lab. Med. 38 (1) (2018) 9–16.
[23] C. Lienhardt, A. Azzurri, A. Amedei, K. Fielding, J. Sillah, O.Y. Sow, et al., Active tuberculosis in Africa is associated with reduced Th1 and increased Th2 activity in vivo, Eur. J. Immunol. 32 (6) (2002) 1605–1613.
[24] Q. Wang, W. Han, J. Niu, B. Sun, W. Dong, G. Li, Prognostic value of serum macrophage migration inhibitory factor levels in pulmonary tuberculosis, Respir.
Res. 20 (1) (2019) 50.
[25] R. Jacobs, E. Tshehla, S. Malherbe, M. Kriel, A.G. Loxton, K. Stanley, et al., Host biomarkers detected in saliva show promise as markers for the diagnosis of pulmonary tuberculosis disease and monitoring of the response to tuberculosis treatment, Cytokine 81 (2016) 50–56.
[26] I.M. Hoel, M.D. Jørstad, M. Ruhwald, T. Mustafa, A.M. Dyrhol-Riise, IP-10 point-of- care tests for monitoring treatment efficacy in tuberculosis in a low-resource setting, Eur. Respir. J. 50 (2017).
[27] I.M. Hoel, M.D. Jørstad, M. Marijani, M. Ruhwald, T. Mustafa, A.M. Dyrhol-Riise, IP-10 dried blood spots assay monitoring treatment efficacy in extrapulmonary tuberculosis in a low-resource setting, Sci. Rep. 9 (1) (2019) 1–9.
[28] M.-R. Lee, C.-J. Tsai, W.-J. Wang, T.-Y. Chuang, C.-M. Yang, L.-Y. Chang, et al., Plasma biomarkers can predict treatment response in tuberculosis patients: a prospective observational study, Medicine 94 (39) (2015).
[29] M.D. Jørstad, M. Marijani, A.M. Dyrhol-Riise, L. Sviland, T. Mustafa, MPT64 antigen detection test improves routine diagnosis of extrapulmonary tuberculosis in a low-resource setting: a study from the tertiary care hospital in Zanzibar, PLoS ONE 13 (5) (2018).
[30] A.K. Niazi, S. Kalra, Diabetes and tuberculosis: a review of the role of optimal glycemic control, J. Diabetes Metab. Disord. 11 (1) (2012) 28.
[31] P. Baghaei, M. Marjani, P. Javanmard, P. Tabarsi, M.R. Masjedi, Diabetes mellitus and tuberculosis facts and controversies, J. Diabetes Metab. Disord. 12 (1) (2013) [32] 58. A. Basit, A. Fawwad, H. Qureshi, A. Shera, Prevalence of diabetes, pre-diabetes and
associated risk factors: second National Diabetes Survey of Pakistan (NDSP), 2016–2017, BMJ Open 8 (8) (2018), e020961.
[33] M.D. Jørstad, A.M. Dyrhol-Riise, J. Aßmus, M. Marijani, L. Sviland, T. Mustafa, Evaluation of treatment response in extrapulmonary tuberculosis in a low-resource setting, BMC Infect. Dis. 19 (1) (2019) 426.
[34] M.R. Purohit, R. Purohit, T. Mustafa, Patient health seeking and diagnostic delay in extrapulmonary tuberculosis: a hospital based study from Central India, Tuberculosis research and treatment. 2019 (2019).
[35] A. Ambreen, M. Jamil, M.A. ur Rahman, T. Mustafa, Viable Mycobacterium tuberculosis in sputum after pulmonary tuberculosis cure, BMC Infect. Dis. 19 (1) (2019) 923.
[36] S. Gupta, V.P. Shenoy, I. Bairy, H. Srinivasa, C. Mukhopadhyay, Diabetes mellitus and HIV as co-morbidities in tuberculosis patients of rural south India, Journal Infect. Public Health 4 (3) (2011) 140–144.
